Samples were processed immediately on an IL-282 CO-oximeter for total hemoglobin, oxyhemoglobin, deoxyhemoglobin, methemoglobin, and carboxy-hemoglobin determinations. No recalibrations were done during the study period in order to establish whether there were tendencies for in vivo SvO to drift over time relative to in vitro SvO,.
All patients enrolled in the study were considered to require hemodynamic monitoring for optimal clinical care. The physicians caring for each patient were given access to the data collected and the information was available to guide therapeutic interventions.
Statistical Analysis
Measurement error was calculated as the difference between in vivo and in vitro SvOt. Correlation coefficients of in vivo vs in vitro Sv02 were also calculated in order to compare current findings with previous studies. Bias (Eq 1), defined as the mean difference between in vivo and in vitro determinations of SvOa, and precision (Eq 2), defined as the SD of the bias, were determined for each patient.
Let i index the n = 10 patients within each group and let j index the n = 21 observations made on each patient. Let SvOa (in vitro^ — in vivOy) denote the difference between the spectrophotometric (COoximetry) measure and the oximetry PA catheter measure for the j measurement in the i* patient.
Bias for the i patient is the average difference between COoximeter and PA catheter values.